Paper Outline: New Observations of Snow Interception from a Subalpine Forest

Author

Alex Cebulski & John Pomeroy

Published

April 10, 2024

Introduction

Forest cover extends over approximately half of North America’s snow-covered region, governing snowpack accumulation and ablation, thereby influencing the timing and magnitude of runoff generation from snowmelt.

  • There is a need for robust models of snow redistribution by vegetation and wind to estimate snow accumulation in mountain forests. To achieve this, a comprehensive understanding of snow redistribution processes is required. However, existing snow interception parameterizations are based on limited observations in distinct climates and forest structures. Rapid changes in climate and forest ecology illustrate the pressing need to assess whether existing snow interception parameterizations are suitable for predicting snow accumulation in diverse and changing environments.

  • Intercepted snow in the canopy is subjected to higher rates of sublimation compared to subcanopy snow due to greater surface area, higher wind speed, and solar exposure (Pomeroy et al., 1998). Across the Northern Hemisphere, researchers estimate that 25 to 45% of annual snowfall may be lost to the sublimation of intercepted snow from the canopy (Essery et al., 2003). Correctly determining the fraction of snowfall intercepted in the canopy is crucial for estimating interception losses by sublimation (Pomeroy et al., 1998). In addition, the time that snow resides in the canopy and is subject to sublimation is dependent on rates of unloading, melt, drip, and resuspension of snow (Hedstrom & Pomeroy, 1998; Katsushima et al., 2023; Lumbrazo et al., 2022; Storck et al., 2002).

  • The theory underpinning current snow interception parameterizations is based on observations ranging from warm maritime (Andreadis et al., 2009; Storck et al., 2002) and cold continental (Ellis et al., 2010; Hedstrom & Pomeroy, 1998; Roesch et al., 2001; Satterlund & Haupt, 1967) climates generally characterized by dense forest canopy. Accurate simulations of forest snow accumulation have been achieved if the parameterizations are applied in similar climates to where they were developed (Lundquist et al., 2021; Rasouli et al., 2019; Roth & Nolin, 2019) or if they are combined into a hybrid parameterization and assessed at global and regional scales in a wide range of climates (Essery et al., 2003; Gelfan et al., 2004). Although accurate performance has been achieved across different climates in some studies (Essery & Pomeroy, 2004; Gelfan et al., 2004), other snow model comparisons (Krinner et al., 2018; Rutter et al., 2009) have shown reduced performance. The decision in earth system models to use parameterizations derived in warm or cold climates is often based on a simple temperature-based step function (Essery et al., 2003; Gelfan et al., 2004) and may require modification to better represent more transitional climates and forest types. The original theory has also been simplified over time, i.e. the increase in canopy coverage with increasing wind speed is not included in more recent parameterizations Roth & Nolin (2019). Updates by Gelfan et al. (2004) to combine the Hedstrom & Pomeroy (1998) and Storck et al. (2002) parameterizations is not typically utilized in recent studies (Krinner et al., 2018; Rutter et al., 2009). The omission or simplified representation of processes and reliance on empirical calibrations likely contribute to model uncertainty when applied in climates and forests where other processes become important (Krinner et al., 2018; Lumbrazo et al., 2022; Lundquist et al., 2021; Moeser et al., 2015; Roth & Nolin, 2019; Rutter et al., 2009).

  • Figure 1 shows the difference in the change in interception efficiency across different snowfall event sizes for three common models. Interception efficiency (interception/snowfall) declines with increasing snow load (Hedstrom & Pomeroy, 1998; Storck et al., 2002) or initially increases and then is followed by a decline (Moeser et al., 2015). The underlying theory of the Moeser et al. (2015) parameterization stems from the Satterlund & Haupt (1967) study who observed an initial increase in the rate of intercepted snow, as snowflakes bridge gaps between needles. It may also be inferred that during the small near 0°C snowfall events observed in Satterlund & Haupt (1967) the majority of snow may have melted immediately due to a warm canopy resulting in low interception efficiency. The initial low interception efficiency was followed by an increase and then flattening off of the interception rate as branches bend due to the weight of snow which Satterlund & Haupt (1967) represented by a numerical analytical sigmoidal function. In the observations by Hedstrom & Pomeroy (1998), snow interception efficiency starts high and then declines. Hedstrom & Pomeroy (1998) hypothesis the shape of this curve is due to a decrease in canopy contact area and change in the incoming snowfall angle of impact as branches bend downward. This relationship was best represented using an inverse exponential function shown in Figure 1. The Hedstrom & Pomeroy (1998) parameterization therefore differs from the (Moeser et al., 2015; Satterlund & Haupt, 1967) sigmoidal function as it does not include a representation for the initially slow interception rate. Andreadis et al. (2009) developed a snow interception model using data collected by Storck et al. (2002) in dense old growth forest in the maritime climate of southwestern Oregon, USA. This method builds off the maximum canopy snow load theory proposed in Hedstrom & Pomeroy (1998) but makes additional modifications to include a step function based on temperature. Here, snow interception efficiency, was found equal to a constant of 0.6 based on snow interception observations from Storck et al. (2002) in southern Oregon.

  • Maximum interception capacity decreases (Hedstrom & Pomeroy, 1998) or increases (Storck et al., 2002) with increasing air temperature (Figure 2). Hedstrom & Pomeroy (1998) proposed that fresh snow density, which may be described as a function of air temperature (Hedstrom & Pomeroy, 1998), plays an important role in governing the interception capacity. Storck et al. (2002) limit, \(W\) as being less than or equal to the maximum interception storage \(W_{max}\) using a step function of temperature based on observations of warmer snow having more cohesion to the canopy.

  • More recent work by Katsushima et al. (2023) collected measurements of snow interception using a weighed tree for a warm-humid coastal environment in Japan. They observed a decline in interception efficiency with increasing wind speed, they attributed to increased hydrometeor velocity and bouncing on impact. While not mentioned in this study, the decrease in interception efficiency may also be due to wind induced unloading. They did not observed a maximum interception capacity within their measurement range of 0-25 mm. Although for temperatures above 0 they could see a decline in interception efficiency above 10 mm maybe due to branch bending + melt rates. Katsushima et al. (2023) suggest air temperature and wind speed alone are insufficient to describe interception efficiency and hypothesize that particle shape may be an improved predictor but did not have the observations to test this. Some of the limited model performance reported by Katsushima et al. (2023) may be attributed to a result of their interception measurements including unloading due to melt and wind.

  • Previous studies have collected measurements of interception efficiency over snowfall events ranging from hourly (Storck et al., 2002) to weekly timesteps (Hedstrom & Pomeroy, 1998). The different measurement time intervals vary in the amount of time possible for ablative processes and may influence model estimates of interception. As a result, some of the interception measurements inevitably include some amount of ablation. Despite the inclusion of unloading in the interception parameterizations developed in these studies, they are often combined with additional unloading parameterization in earth system models (Clark et al., 2015; Ellis et al., 2013) leading to some potential of double counting of the ablation process.

  • Uncertainties also arise in the scaling of point or branch scale measurements to the plot scale (Staines & Pomeroy, 2023).

  • Forest structure governs the interception efficiency observed at a given location (Hedstrom & Pomeroy, 1998; Roth & Nolin, 2019). Metrics used in common snow interception parameterizations (Hedstrom & Pomeroy, 1998; Storck et al., 2002) to describe forest structure include canopy cover and leaf area index (LAI). Leaf area index is defined by Chen et al. (1997) as one half the total green leaf area per unit ground surface area. Canopy cover is defined in Hedstrom & Pomeroy (1998) as the fraction of sky not visible by the instrument from under the canopy. While more detailed forest structure metrics exist derived from detailed LiDAR scans (Helbig et al., 2020; Roth & Nolin, 2019), often they are not available at regional extents required to run hydrological models.

  • Add sentence on Staines & Pomeroy (2023).

  • Several processes govern the accumulation of snow in mountain forests, and the importance of individual processes may differ depending on climate and forest structure (Gelfan et al., 2004; Hedstrom & Pomeroy, 1998; Moeser et al., 2015; Staines & Pomeroy, 2023). Therefore determining the dominant processes in varying climate and forests structures is important to guide model decision makers on what existing model parameterizations to choose or if a hybrid approach may be appropriate.

  • New observations of snow interception and ablation processes will help determine if existing theories are applicable in differing climates and diverse forest structures.

Figure 1: Example simulation of snow interception parameterizations by Hedstrom and Pomeroy (1998) (HP98), combined Storck et al. (2002) and Andreadis et al., (2009) (SA09), and the Moeser et al., (2009) (M15). Event interception efficiency is the change in canopy SWE storage divided by the corresponding change in SWE in the open. The temperature for the example events are held constant at -5 °C. The canopy coverage values are 0.75 (low), medium (0.80) and high (0.83).
Figure 2: A comparison of the Hedstrom & Pomeroy (1998) (HP98) and Andreadis et al. (2009) (SA09) interception capacity parameterizations. Since the HP98 parameterization is a function of new snow density as a result the interception capacity is negatively related to air temperature.
  • The novelty of this study is the study site location in a windswept discontinuous subalpine ridge forest and the use of high temporal frequency automated measurements and discrete high spatial resolution measurements using aerial LiDAR to attempt to separate out interception from ablative processes. Timelapse cameras were also used to confirm absence of unloading during interception periods.

Objective: To assess the influence of canopy structure and meteorology on snow interception processes in a windswept subalpine forest.

Research Questions:

  1. Are the theories and assumptions of existing snow interception parameterizations true for field measurements collected in a continental subalpine forest?
  2. What are the dominant processes that control snow interception in a subalpine forest and are these included in existing parameterizations?

Specific Questions

  • Is the influence of meteorology important for governing I/P?
    • have not found strong association with met vars at one location but when looking across the forest wind appears to influence hydrometeor contact angle…
  • What is the influence of wind governing hydrometeor contact angle in sparse forest?
    • need to determine if the association between contact probability and I/P has an angular dependance (i.e., show hemisphere of R2 between contact probability and I/P as in Staines & Pomeroy (2023))
    • is this association stronger when there is snow in the canopy i.e., branches have been bridged or reduced due to branch bending?

Theory

Figure 3: The mass balance of intercepted snow in a coniferous forest canopy and the subcanopy snowpack. The colours of the arrows correspond to the water phase: solid (purple), liquid (blue) and vapour (light green). The head of the arrow indicates a positive flux either into the control volume (positive) or away from the control volume (negative). Note that the fluxes may transition between positive and negative. In the case of sublimation, from the canopy or snowpack, the flux may be positive (sublimation) or negative (deposition). This figure was adapted from Pomeroy and Gray, (1995).

Figure 3 shows the mass exchange of snowfall between the forest canopy and the snowpack on the ground surface. The storage of snow water equivalent (SWE) is represented here with respect to the canopy (\(W\), mm) or the surface snowpack (\(S\), mm). Fluxes that are repeated between the canopy and snowpack control volume have a superscript to specify what control volume they refer to (i.e. \(q^{veg}\) refers to the vegetation control volume and \(q^{snow}\) refers to the surface snowpack control volume).

The change in canopy SWE storage, \(W\) (mm), may be represented as:

\[ \frac{dW}{dt} = q_{sf} - q_{tf}(W) - q_{unld}(W) - q_{drip}(W) - q_{wind}^{veg}(W) - q_{sub}^{veg}(W) \tag{1}\]

where \(q_{sf}\) (mm s-1) is the above canopy snowfall rate, \(q_{tf}\) (mm s-1) is the throughfall rate, \(q_{wind}^{veg}\) (mm s-1) is the wind transport rate in our out of the control volume (typically assumed to be negligible in the literature), \(q_{sub}^{veg}\) (mm s-1) is the intercepted snow sublimation rate, \(q_{unld}\) (mm s-1) is the canopy snow unloading rate and \(q_{drip}\) (mm s-1) is the canopy snow drip rate due to canopy snowmelt. \(q_{wind}^{veg}\) and \(q_{sub}^{veg}\) may be a positive or negative flux. Where all of the above rates are a function of snow load (\(W\)), which is how much snow is present in the canopy.

Hedstrom & Pomeroy (1998) hypothesized that a change in hydrometeor trajectory angle was a factor in influencing interception efficiency. Figure 4 shows the theoretical increase in trajectory angle with increasing wind speed.

Figure 4: The theoretical relationship of hydrometeor trajectory (departure from horizontal plane) with increasing wind speed. This assumes a constant hydrometor velocity of 1 m/s, typical of snowfall and a horizontal velocity equal to wind speed.

The trajectory angle, \(\theta_h\) of a hydrometeor as the departure in degrees (°) from a horizontal plane (i.e., 90° for vertical snowfall), may be calculated as:

\[ \theta_h = \arctan \left(\frac{-rise}{run}\right)*\frac{180}{\pi} \tag{2}\]

where \(rise\) is the terminal fall velocity of the hydrometeor (m s-1), \(run\) is the horizontal change in the hydrometeor (m s-1) and is assumed equal to the wind speed.

Methods

Study Site

This study was conducted at the Subalpine Ridge Forest (SRF) site located in the Fortress Mountain Research Basin (FMRB), AB ~ 51 ◦ N, 2100 m asl., a continental headwater basin with sparse subalpine forest (Figure 5 (a)). In-situ observations of forest snow accumulation, snow interception, and canopy snow ablation were collected over the 2022 and 2023 water years. The species of trees at the SRF site is comprised of coexisting subalpine fir (Abies lasiocarpa) and Engelmann spruce (Picea engelmannii), with a proportion of 70% and 30% respectively (Langs et al., 2020). In the early 1900s the majority of the SRF vegetation burned during a large forest fire that affected most of the Kananaskis Valley. Following the fire, the forest has naturally regenerated with minimal disturbance.

A 15 m tall flux tower provided measurements within the SRF site provided measurements of the meteorological conditions at 15-min time intervals (Figure 5 (b)). The meteorological variables used in this study include air temperature, relative humidity and wind speed (4.5 m above the ground). A weighing precipitation gauge (OTT Pluvio) provided measurements of the snowfall rate (\(q_{sf}\)) to an open clearing adjacent to the study area shown in Figure 5 (b). The snowfall rate was corrected for undercatch following phase correction by Harder & Pomeroy (2013) and catch efficiency by Macdonald & Pomeroy (2007).

(a)
(b)
Figure 5: Regional map showing location of Fortress Mountain (a) and map of forest plots, flux towers, and survey transects (b) in this study.

Snow Surveys

Areal snow surveys provided measurements of throughfall over a discrete time interval (\(\Delta {tf}\)). Twelve snow surveys (pre and post event pairs) were selected from a total of 39 snow surveys, resulting in six measurements of snow interception efficiency over the stations shown in Figure 5 (b). If a pre event crust layer was present, post event measurements of fresh snow accumulation above the crust layer were interpreted as throughfall over the event. In the absence of a defined crust layer, the difference of pre and post event snow depth to ground was interpreted as event throughfall. An average of 40 snow depth samples were taken for each snow survey with forest structure ranging from canopy clearing to dense canopy (Figure 5 (b)). A 1000 cc snow density wedge sampler was used to measure the density of the fresh snow layer. The duration of each event, shown in Table 1, ranged from 26 to 148 hours.

Lysimeter Data

Three subcanopy lysimeters (e.g., Figure 6 (a)) provided fifteen minute interval measurements of throughfall plus unloading (instrument locations are shown in Figure 5 (b)). For select events where ablative processes, \(q_{unld}(W)\), \(q_{drip}\) and \(q_{wind}^{veg}\) could be considered negligible, the subcanopy lysimeters were inferred to provide measurements of throughfall. Timelapse imagery, a weighed tree and in-situ observations were used to ensure ablation of snow intercepted in the canopy or snow on the ground was minimal over each of the selected events.

A weighed tree lysimeter, shown in Figure 6 (b) and Figure 6 (c), measured the weight of canopy snow load, \(W_{wt}\) (kg). A live subalpine fir (Abies lasiocarpa) tree was cut and suspended from a load cell at the beginning of the 2022 and 2023 water years which recorded the weight of the tree. The bottom of the tree was sealed to limit some transpiration and to prevent spinning and abrupt impacts of the free-hanging tree, the base of the tree was attached to a support system that allows for vertical movement but limits abrupt horizontal movements. The weight of snow in the canopy was scaled to an areal estimate of canopy snow load (\(W\), kg m-2) using manual snow survey measurements (as in Hedstrom & Pomeroy, 1998).

Snow Interception

Throughfall measurements collected form manual snow surveys (\(q_{tf}\cdot \Delta t\)) and the subcanopy lysimeters (\(q_{tf}\)) were used to estimate the amount of snow intercepted in the canopy. During calm snowfall periods Equation 1 was be simplified to estimate the amount of snow intercepted in the canopy:

\[ \frac{dW}{dt} = q_{sf}-q_{tf}(W) \tag{3}\]

This method was preferred, compared to measurements from the weighed tree lysimeter, as the subcanopy lysimeters were not influenced by sublimation losses from snow intercepted in the canopy.

Interception efficiency, \(\frac{I}{P}\) (-), which is the fraction of snow intercepted over a discrete time interval, \(\Delta t\) was calculated as:

\[ \frac{I}{P} = \frac{\Delta W}{\overline{q_{sf}} \Delta t} \tag{4}\]

where \(\Delta W\) (mm) is the increase in canopy load over a discrete time interval and \(\overline{q_{sf}}\) (mm), is the average snowfall rate over \(\Delta t\).

Forest Structure Metrics

Upward-facing hemispherical photographs were analyzed to provide measurements of LAI and canopy coverage for each snow survey station and lysimeter location. The camera used was a Nikon Coolpix 4500 with a EC-F8 hemispherical lens. The images were cropped and automatically thresholded to differentiate between sky and canopy using the hemispheR R package (Chianucci & Macek, 2023). Forest metrics were produced for vertical zenith angles (VZA) 15°, 30°, 45° and 60°.

Table 1: Meteorological conditions and interception observations for six selected events that coincide with pre and post snow surveys. Snowfall (mm) and ‘Canopy Snow Load (mm)’ are the cumulative respective snowfall rate and interception rate over each event. Air Temperature (°C), Relative Humidity (%), Wind Speed (m/s) are the event mean. Event I/P is calculated as Snowfall (mm) divided by Canopy Snow Load (mm).
Start Date Event Duration (hours) Snowfall (mm) Canopy Snow Load (mm) Event I/P (-) Air Temperature (°C) Relative Humidity (%) Wind Speed (m/s)
2022-03-31 145.0 hours 22.558421 10.572707 0.4686812 -4.00 60.49 1.69
2022-04-08 128.0 hours 16.322948 6.353680 0.3892483 -9.87 67.33 1.16
2023-01-26 28.0 hours 9.846485 4.039235 0.4102210 -2.67 72.81 1.30
2023-02-15 148.0 hours 32.298464 14.493226 0.4487280 -8.74 70.38 1.63
2023-03-13 26.5 hours 28.370579 10.272936 0.3620982 -3.90 86.12 1.34
2023-03-24 26.0 hours 26.811832 13.486138 0.5029920 -6.10 89.56 0.69
(a)
(b)
(c)
Figure 6: Subcanopy lysimeter (a), weighed tree lysimeter loaded with snow (b) and weighed tree lysimeter bare of snow (c).

Results

Subcanopy Lysimeters

A linear increase in canopy storage with increasing snowfall, is shown in Figure 7 (b) for 25 snowfall events. The difference in the rate of change between the events, shown in Figure 7 (b), does not show a strong connection between cold events compared to warmer events. However, this change in slope is interpreted to be a result of a combination of meteorological variables which will be explored in the following section.

(a)
(b)
Figure 7: Cumulative canopy snow storage measured using the subcanopy lysimeters plotted against cumulative event snowfall for five selected snowfall events. The colour of each line shows the average air temperature (a) and wind speed (b) over the duration of each event.

Air temperature was observed to have little influence over interception efficiency measured using the subcanopy lysimeters and snowfall gauge for the 25 snowfall events (Figure 8, A & B). Relative humidity was observed to have a slight negative association with interception efficiency. However this is primarily based on the lowest relative humidity bin with relatively few observations. Wind speed was observed to increase the mean interception efficiency slightly from 0.6 to 0.63 between wind speed bins of 0.25 and 0.75 m s-1 (Figure 8, C). This is thought to be due to an associated increase in canopy contact area. The mean interception efficiency decreases wind speeds above 2 m s-1 to a minimum of 0.47 for the 3.75 m s-1 wind speed bin. The initial canopy snow load had a positive association with interception efficiency as the canopy filled with snow, increasing the mean interception efficiency from 0.57 to 0.67. Followed by a gradual decline in interception efficiency for snow loads greater than 7.5 mm to a minimum of 0.48 at snow loads above 18 mm (Figure 8, C). As the canopy fills with snow, interception efficiency was expected to increase due to snow bridging across the branches increasing the contact area of the canopy. Interception efficiency was also expected to decline at higher snow loads due some increase in unloading due to branch bending and a decline in the canopy contact area as branches bend down. Hydrometeor diameter and velocity did not have a strong association with interception efficiency.

Figure 8: Scatter plots of discrete observations (green) of snow interception efficiency observed at 15 minute intervals using the subcanopy lysimeter and snowfall gauge against and binned data (black). Panels show (A) air temperature, (B) wind speed, (C) initial canopy snow load (the snow load observed at the begining of the timestep), (E) hydrometeor diameter, (F) hydrometeor velocity. The black open circles show the mean of each bin and the error bars represent the standard deviations. The data were filtered to include observations with a snowfall rate > 0 mm/hr and a snowfall rate > the subcanopy lysimeter throughfall rate to minimize observations with unloading. Periods of unloading and melt were also removed through careful analysis of the weighed tree, subcanopy lysimeters, and timelapse imagery.

Hydrometeor Size and Velocity

No relationship between wind speed and hydrometeor velocity was found between fifteen minute average measurements over the study period (Figure 9). However, above wind speeds of 2 m s-1 an increase in velocity was observed which was also associated with smaller hydrometeor diameters.

Figure 9: Scatter plot illustrating the relationship of wind speed and hydrometeor velocity. The dots represent the 15 minute averages of both wind speed and hydrometeor velocity coloured by their diameter (mm). The boxplots overlayed on top of the dots show the median (50th percentile), interquartile range (IQR) (25-75 percentile), smallest and largest value within 1.5 * the IQR.

Aerial Lidar

The mean contact number was calculated for a snow-free canopy across each azimuth of the hemisphere at a 1 deg resolution and then averaged for each zenith angle for each forest plot to determine how canopy contact number is associated with increasing hydrometeor trajectory. Figure 10 shows an exponential increase in the number of canopy contacts as the trajectory becomes more horizontal (near 0 °). Forest plot PWL_SW which has a more closed canopy has a higher mean contact number at near vertical trajectories (near -90°) and increases at a slower rate compared to sparser canopies like FSR_S which start off with lower mean contact number and increase at a higher rate.

Figure 10: Scatter plot showing the association of hydrometeor trajectory with mean contact number. The dots represent the mean contact number across the hemisphere for a trajectory (zenith angle) at each forest plot. The black lines show a log transformed linear regression (solid black line) and the nonlinear-least-squares regress (dashed black line).

TODO: maybe include a second plot for the snow covered day

Aerial lidar measurements of snow accumulation illustrate how hydrometeor trajectory influences effective canopy coverage and thus, interception efficiency. A 24 hr long snowfall accumulation event on March 13, 2023 was characterized by a median hydrometeor velocity of 0.9 m s-1, wind speed of 1.4 m s-1 and wind direction of 183° which resulting in a predicted average hydrometeor trajectory of -33° from 183° (South).

For a more closed canopy (Figure 11 (a), PWL_E), the correlation between interception efficiency and mean contact number across the hemisphere, agreegated for each grid cell, within each plot, shows that the highest Spearman’s R correlation is observed between zenith angles of 0-30° and azimuth of 180-190° South. However, for more open canopy (Figure 11 (b), FSR_S) the correlation is lower and the effect is not quite as localised around 0-30° and azimuth of 180-190°.

(a) PWL E
(b) FSR S
Figure 11: Spearman’s Correlation of contact number and interception efficiency for Powerline South (a) and Subalpine Ridge Forest (b).

Based on the zone of high correlation in Figure 11 (a) and the predicted hydrometeor trajectory using Equation 2, the hemisphere was filtered to 13-16 deg and azimuth of 180-190 to assess the correlation between mean contact number and interception efficiency. Figure 12 (a) shows a logarithmic relationship between interception efficiency and mean contact number for the PWL E plot. Figure 12 (b) shows a slightly more logarithmic relationship for the FSR S plot.

(a) PWL E
(b) FSR S
Figure 12: Scatterplot of contact number and interception efficiency using zenith angles of 13-16 and azimuth of 180-190 for Powerline South (a) and Subalpine Ridge Forest (b).

TODO: decide on using estimated return number, contact number, or transmittance here.

event_id Air Temp. (°C) RH (%) Wind Dir. (°) Wind Speed (m/s) Cuml. Snowfall (mm)
2022-02-16 -7.020461 88.14542 188.60 0.4709743 4.110716
2023-01-27 -3.407188 77.06561 149.95 1.0850000 9.632316
2023-03-14 -3.476119 86.76766 183.40 1.3921543 28.370579

Discussion

  • Forest structure is the primary control in determining the fraction of snowfall intercepted at a given location. A low range in interception efficiency was observed across events with varying snowfall, air temperature, relative humidity, hydrometeor diameter and hydrometeor velocity.

  • Empirical evidence to support the theory of increased in canopy contact area with increased wind speed (Hedstrom & Pomeroy, 1998) has not been provided in the literature. The snow survey measurements conducted here corroborated with this theory, showing that events with high wind resulted in an increase in interception efficiency at sparse canopy locations. High frequency measurements by the subcanopy lysimeters also showed a slight increase in interception efficiency with increasing wind speed, but at peak wind speeds interception efficiency decreased due to unloading.

  • Existing theory suggests either a positive (Storck et al., 2002) or negative (Hedstrom & Pomeroy, 1998) association of air temperature with the maximum canopy storage (Figure 2). The findings presented in this study deviate from the relationships outlined in Hedstrom & Pomeroy (1998) and Storck et al. (2002) concerning the association between air temperature and \(W_{max}\). Specifically, the findings suggest an absence of relationship between air temperature and interception efficiency. The increase in canopy storage with snowfall also did not show any dependence on air temperature or a plateau in storage suggesting the maximum canopy storage value may not have been reached and is likely greater than 25 mm.

  • Satterlund & Haupt (1967) showed that interception efficiency increases as the canopy fills with snow, while later declining due to branch bending. Hedstrom & Pomeroy (1998) and Storck et al. (2002) did not observe this initial increase. Observations from the subcanopy lysimeters here showed some evidence for an initial increase in interception efficiency with increasing canopy snow load followed by a decline in interception efficiency likely due to unloading or reduced canopy coverage. The decline in interception efficiency at high canopy snow loads was at rate much slower than has been observed by previous studies (Hedstrom & Pomeroy, 1998; Moeser et al., 2015; Satterlund & Haupt, 1967). At higher canopy snow loads, several studies suggest interception efficiency declines (Hedstrom & Pomeroy, 1998; Moeser et al., 2015; Satterlund & Haupt, 1967), based on the premise of reduced canopy coverage due and catch efficiency to branch bending. However, the strong exponential decline in the interception efficiency observed with increasing event snowfall shown in Figure 1 may be a result of increased unloading rates as branches bend down. The long duration of snowfall events included in Hedstrom & Pomeroy (1998) also increase the likely hood of unloading. This potential inclusion of unloading within the interception parameterizations provided in (Hedstrom & Pomeroy, 1998; Moeser et al., 2015; Satterlund & Haupt, 1967) may lead to double counting of unloading when combined with an additional unloading parameterization as in (Hedstrom & Pomeroy, 1998). However, if interception efficiency is known to decline due to an associated decrease in canopy coverage from branch bending it may be appropriate to have some decline in interception efficiency at higher canopy loads while also having a separate parameterization to increase unloading with canopy snow load.

  • Little increase in hydrometeor velocity was observed with increaseing wind speed which supported the use of a constant terminal fall velocity in Equation 2. The increase in velocity above 2 m/s may be due to smaller diameter hydrometeors associated with convective storm activity.

Conclusions

  • The forest structure is the main factor in governing the fraction of intercepted snowfall at a particular site, with meteorological conditions contributing less to variability.
  • Forest structure metrics calculated using high zenith angles better described the variability in interception efficiency in this windswept subalpine forest.
  • Interception efficiency increased with increasing canopy coverage and LAI. However, the strength of this association was reduced at higher wind speeds.
  • High wind speeds increased interception efficiency due to an associated increase in canopy contact area for sparsely forested locations, while later decreasing intercepted load due to increased snow unloading.
  • No influence of air temperature on interception efficiency or maximum canopy storage was observed.
  • Interception efficiency increased slightly as the canopy filled with snow and declined later at higher loads.
  • Unloading increased with increasing canopy storage.
  • The maximum canopy storage capacity for this study site is likely higher than was observed here.

References

Andreadis, K. M., Storck, P., & Lettenmaier, D. P. (2009). Modeling snow accumulation and ablation processes in forested environments. Water Resources Research, 45(5), 1–33. https://doi.org/10.1029/2008WR007042
Chen, J. M., Rich, P. M., Gower, S. T., Norman, J. M., & Plummer, S. (1997). Leaf area index of boreal forests: Theory, techniques, and measurements. Journal of Geophysical Research Atmospheres, 102(24), 29429–29443. https://doi.org/10.1029/97jd01107
Chianucci, F., & Macek, M. (2023). hemispheR: an R package for fisheye canopy image analysis. Agricultural and Forest Meteorology. https://doi.org/10.1016/j.agrformet.2023.109470
Clark, M. P., Nijssen, B., Lundquist, J. D., Kavetski, D., Rupp, D. E., Woods, R. A., Freer, J. E., Gutmann, E. D., Wood, A. W., Brekke, L. D., Arnold, J. R., Gochis, D. J., & Rasmussen, R. M. (2015). A unified approach for process-based hydrologic modeling: 1. Modeling concept. Water Resources Research, 51(4), 2498–2514. https://doi.org/https://doi.org/10.1002/2015WR017198
Ellis, C. R., Pomeroy, J. W., Brown, T., & MacDonald, J. (2010). Simulation of snow accumulation and melt in needleleaf forest environments. Hydrology and Earth System Sciences, 14(6), 925–940. https://doi.org/10.5194/hess-14-925-2010
Ellis, C. R., Pomeroy, J. W., & Link, T. E. (2013). Modeling increases in snowmelt yield and desynchronization resulting from forest gap-thinning treatments in a northern mountain headwater basin. Water Resources Research, 49(2), 936–949. https://doi.org/10.1002/wrcr.20089
Essery, R., & Pomeroy, J. W. (2004). Vegetation and topographic control of wind-blown snow distributions in distributed and aggregated simulations for an arctic tundra basin. Journal of Hydrometeorology, 5(5), 735–744. https://doi.org/10.1175/1525-7541(2004)005<0735:VATCOW>2.0.CO;2
Essery, R., Pomeroy, J. W., Parviainen, J., & Storck, P. (2003). Sublimation of snow from coniferous forests in a climate model. Journal of Climate, 16(11), 1855–1864. https://doi.org/10.1175/1520-0442(2003)016<1855:SOSFCF>2.0.CO;2
Gelfan, A. N., Pomeroy, J. W., & Kuchment, L. S. (2004). Modeling forest cover influences on snow accumulation, sublimation, and melt. Journal of Hydrometeorology, 5(5), 785–803. https://doi.org/10.1175/1525-7541(2004)005<0785:MFCIOS>2.0.CO;2
Harder, P., & Pomeroy, J. W. (2013). Estimating precipitation phase using a psychrometric energy balance method. Hydrological Processes, 27(13), 1901–1914.
Hedstrom, N. R., & Pomeroy, J. W. (1998). Measurements and modelling of snow interception in the boreal forest. Hydrological Processes, 12(10-11), 1611–1625. https://doi.org/10.1002/(SICI)1099-1085(199808/09)12:10/11<1611::AID-HYP684>3.0.CO;2-4
Helbig, N., Moeser, D., Teich, M., Vincent, L., Lejeune, Y., Sicart, J. E. J.-E. J. E., & Monnet, J. M. J.-M. (2020). Snow processes in mountain forests: interception modeling for coarse-scale applications. Hydrology and Earth System Sciences, 24(5), 2545–2560. https://doi.org/10.5194/hess-24-2545-2020
Katsushima, T., Kato, A., Aiura, H., Nanko, K., Suzuki, S., Takeuchi, Y., & Murakami, S. (2023). Modelling of snow interception on a Japanese cedar canopy based on weighing tree experiment in a warm winter region. Hydrological Processes, 37(6), 1–16. https://doi.org/10.1002/hyp.14922
Krinner, G., Derksen, C., Essery, R., Flanner, M., Hagemann, S., Clark, M. P., Hall, A., Rott, H., Brutel-Vuilmet, C., Kim, H., Ménard, C. B., Mudryk, L., Thackeray, C., Wang, L., Arduini, G., Balsamo, G., Bartlett, P., Boike, J., Boone, A., … Zhu, D. (2018). ESM-SnowMIP: Assessing snow models and quantifying snow-related climate feedbacks. Geoscientific Model Development, 11(12), 5027–5049. https://doi.org/10.5194/gmd-11-5027-2018
Langs, L. E., Petrone, R. M., & Pomeroy, J. W. (2020). A \(\delta\)18O and \(\delta\)2H stable water isotope analysis of subalpine forest water sources under seasonal and hydrological stress in the Canadian Rocky Mountains. Hydrological Processes, 34(26), 5642–5658. https://doi.org/10.1002/hyp.13986
Lumbrazo, C., Bennett, A., Currier, W. R., Nijssen, B., & Lundquist, J. (2022). Evaluating Multiple Canopy-Snow Unloading Parameterizations in SUMMA With Time-Lapse Photography Characterized by Citizen Scientists. Water Resources Research, 58(6), 1–22. https://doi.org/10.1029/2021WR030852
Lundquist, J. D., Dickerson-Lange, S., Gutmann, E., Jonas, T., Lumbrazo, C., & Reynolds, D. (2021). Snow interception modelling: Isolated observations have led to many land surface models lacking appropriate temperature sensitivities. Hydrological Processes, 35(7), 1–20. https://doi.org/10.1002/hyp.14274
Macdonald, J., & Pomeroy, J. W. (2007). Gauge Undercatch of Two Common Snowfall Gauges in a Prairie Environment. Proceedings of the 64th Eastern Snow Conference, St. John‘s, Canada., 1974, 119–126.
Moeser, D., Stähli, M., & Jonas, T. (2015). Improved snow interception modeling using canopy parameters derived from airborne LiDAR data. Water Resources Research, 51(7), 5041–5059.
Pomeroy, J. W., Parviainen, J., Hedstrom, N., & Gray, D. M. (1998). Coupled modelling of forest snow interception and sublimation. Hydrological Processes, 12(15), 2317–2337. https://doi.org/10.1002/(SICI)1099-1085(199812)12:15<2317::AID-HYP799>3.0.CO;2-X
Rasouli, K., Pomeroy, J. W., & Whitfield, P. H. (2019). Are the effects of vegetation and soil changes as important as climate change impacts on hydrological processes? Hydrology and Earth System Sciences, 23(12), 4933–4954. https://doi.org/10.5194/hess-23-4933-2019
Roesch, A., Wild, M., Gilgen, H., & Ohmura, A. (2001). A new snow cover fraction parameterization for the ECHAM4 GCM. Climate Dynamics, 17(12), 933–946. https://doi.org/10.1007/s003820100153
Roth, T. R., & Nolin, A. W. (2019). Characterizing Maritime Snow Canopy Interception in Forested Mountains. Water Resources Research, 55(6), 4564–4581. https://doi.org/10.1029/2018WR024089
Rutter, N., Essery, R., Pomeroy, J. W., Altimir, N., Andreadis, K. M., Baker, I., Barr, A., Bartlett, P., Boone, A., Deng, H., Douville, H., Dutra, E., Elder, K., Ellis, C. R., Feng, X., Gelfan, A., Goodbody, A., Gusev, Y., Gustafsson, D., … Yamazaki, T. (2009). Evaluation of forest snow processes models (SnowMIP2). Journal of Geophysical Research: Atmospheres, 114(D6), 10–18. https://doi.org/10.1029/2008JD011063
Satterlund, D. R., & Haupt, H. F. (1967). Snow catch by Conifer Crowns. Water Resources Research, 3(4), 1035–1039.
Staines, J., & Pomeroy, J. W. (2023). Influence of forest canopy structure and wind flow on patterns of sub-canopy snow accumulation in montane needleleaf forests. Hydrological Processes, 37(10), 1–19. https://doi.org/10.1002/hyp.15005
Storck, P., Lettenmaier, D. P., & Bolton, S. M. (2002). Measurement of snow interception and canopy effects on snow accumulation and melt in a mountainous maritime climate, Oregon, United States. Water Resources Research, 38(11), 1–16. https://doi.org/10.1029/2002wr001281